Because of promising leads provided by the literature and our pilot data, we feel it is essential to continue to gather more comprehensive information regarding the impact of autism treatments on both child and family measures in order to more fully evaluate the effects of the treatment packages. This will involve a continuation of the measures we obtained during the first years of our project. In addition, we propose to pursue promising pilot data on "pivotal" target behaviors. The pilot data thus far collected suggest that identifying and treating certain behaviors holds promise for constructing new packages, affecting a wide range of clinically significant behaviors. We suspect that two basic packages will emerge. One package will focus on variables relating to generalization. In this package we anticipate including programs to increase spontaneity, utilizing observational learning and inconsistent contingent consequences. The second major package will focus on variables related to increasing responsivity and will include programs to increase responsiveness to multiple cues and subsequently enhance social responsiveness and incidental learning. Thus, in the course of the next three years we propose to use comprehensive assessments, in the context of both within subject and group designs, to evaluate these additional packages. Finally, as before, we will integrate these new packages into our ongoing parent training program. This should help ensure that the parents acquire the skills and curriculum that will maximize the breadth, generality and maintenance of treatment gains in their children.